Coverage Report

Created: 2026-05-16 06:52

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/gmp/mpn/tdiv_qr.c
Line
Count
Source
1
/* mpn_tdiv_qr -- Divide the numerator (np,nn) by the denominator (dp,dn) and
2
   write the nn-dn+1 quotient limbs at qp and the dn remainder limbs at rp.  If
3
   qxn is non-zero, generate that many fraction limbs and append them after the
4
   other quotient limbs, and update the remainder accordingly.  The input
5
   operands are unaffected.
6
7
   Preconditions:
8
   1. The most significant limb of the divisor must be non-zero.
9
   2. nn >= dn, even if qxn is non-zero.  (??? relax this ???)
10
11
   The time complexity of this is O(qn*qn+M(dn,qn)), where M(m,n) is the time
12
   complexity of multiplication.
13
14
Copyright 1997, 2000-2002, 2005, 2009, 2015 Free Software Foundation, Inc.
15
16
This file is part of the GNU MP Library.
17
18
The GNU MP Library is free software; you can redistribute it and/or modify
19
it under the terms of either:
20
21
  * the GNU Lesser General Public License as published by the Free
22
    Software Foundation; either version 3 of the License, or (at your
23
    option) any later version.
24
25
or
26
27
  * the GNU General Public License as published by the Free Software
28
    Foundation; either version 2 of the License, or (at your option) any
29
    later version.
30
31
or both in parallel, as here.
32
33
The GNU MP Library is distributed in the hope that it will be useful, but
34
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
35
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
36
for more details.
37
38
You should have received copies of the GNU General Public License and the
39
GNU Lesser General Public License along with the GNU MP Library.  If not,
40
see https://www.gnu.org/licenses/.  */
41
42
#include "gmp-impl.h"
43
#include "longlong.h"
44
45
46
void
47
mpn_tdiv_qr (mp_ptr qp, mp_ptr rp, mp_size_t qxn,
48
       mp_srcptr np, mp_size_t nn, mp_srcptr dp, mp_size_t dn)
49
110k
{
50
110k
  ASSERT_ALWAYS (qxn == 0);
51
52
110k
  ASSERT (nn >= 0);
53
110k
  ASSERT (dn >= 0);
54
110k
  ASSERT (dn == 0 || dp[dn - 1] != 0);
55
110k
  ASSERT (! MPN_OVERLAP_P (qp, nn - dn + 1 + qxn, np, nn));
56
110k
  ASSERT (! MPN_OVERLAP_P (qp, nn - dn + 1 + qxn, dp, dn));
57
58
110k
  switch (dn)
59
110k
    {
60
0
    case 0:
61
0
      DIVIDE_BY_ZERO;
62
63
8.22k
    case 1:
64
8.22k
      {
65
8.22k
  rp[0] = mpn_divrem_1 (qp, (mp_size_t) 0, np, nn, dp[0]);
66
8.22k
  return;
67
0
      }
68
69
3.83k
    case 2:
70
3.83k
      {
71
3.83k
  mp_ptr n2p;
72
3.83k
  mp_limb_t qhl, cy;
73
3.83k
  TMP_DECL;
74
3.83k
  TMP_MARK;
75
3.83k
  if ((dp[1] & GMP_NUMB_HIGHBIT) == 0)
76
3.03k
    {
77
3.03k
      int cnt;
78
3.03k
      mp_limb_t d2p[2];
79
3.03k
      count_leading_zeros (cnt, dp[1]);
80
3.03k
      cnt -= GMP_NAIL_BITS;
81
3.03k
      d2p[1] = (dp[1] << cnt) | (dp[0] >> (GMP_NUMB_BITS - cnt));
82
3.03k
      d2p[0] = (dp[0] << cnt) & GMP_NUMB_MASK;
83
3.03k
      n2p = TMP_ALLOC_LIMBS (nn + 1);
84
3.03k
      cy = mpn_lshift (n2p, np, nn, cnt);
85
3.03k
      n2p[nn] = cy;
86
3.03k
      qhl = mpn_divrem_2 (qp, 0L, n2p, nn + (cy != 0), d2p);
87
3.03k
      if (cy == 0)
88
1.14k
        qp[nn - 2] = qhl; /* always store nn-2+1 quotient limbs */
89
3.03k
      rp[0] = (n2p[0] >> cnt)
90
3.03k
        | ((n2p[1] << (GMP_NUMB_BITS - cnt)) & GMP_NUMB_MASK);
91
3.03k
      rp[1] = (n2p[1] >> cnt);
92
3.03k
    }
93
797
  else
94
797
    {
95
797
      n2p = TMP_ALLOC_LIMBS (nn);
96
797
      MPN_COPY (n2p, np, nn);
97
797
      qhl = mpn_divrem_2 (qp, 0L, n2p, nn, dp);
98
797
      qp[nn - 2] = qhl; /* always store nn-2+1 quotient limbs */
99
797
      rp[0] = n2p[0];
100
797
      rp[1] = n2p[1];
101
797
    }
102
3.83k
  TMP_FREE;
103
3.83k
  return;
104
0
      }
105
106
98.4k
    default:
107
98.4k
      {
108
98.4k
  int adjust;
109
98.4k
  gmp_pi1_t dinv;
110
98.4k
  TMP_DECL;
111
98.4k
  TMP_MARK;
112
98.4k
  adjust = np[nn - 1] >= dp[dn - 1];  /* conservative tests for quotient size */
113
98.4k
  if (nn + adjust >= 2 * dn)
114
61.9k
    {
115
61.9k
      mp_ptr n2p, d2p;
116
61.9k
      mp_limb_t cy;
117
61.9k
      int cnt;
118
119
61.9k
      qp[nn - dn] = 0;        /* zero high quotient limb */
120
61.9k
      if ((dp[dn - 1] & GMP_NUMB_HIGHBIT) == 0) /* normalize divisor */
121
5.15k
        {
122
5.15k
    count_leading_zeros (cnt, dp[dn - 1]);
123
5.15k
    cnt -= GMP_NAIL_BITS;
124
5.15k
    d2p = TMP_ALLOC_LIMBS (dn);
125
5.15k
    mpn_lshift (d2p, dp, dn, cnt);
126
5.15k
    n2p = TMP_ALLOC_LIMBS (nn + 1);
127
5.15k
    cy = mpn_lshift (n2p, np, nn, cnt);
128
5.15k
    n2p[nn] = cy;
129
5.15k
    nn += adjust;
130
5.15k
        }
131
56.8k
      else
132
56.8k
        {
133
56.8k
    cnt = 0;
134
56.8k
    d2p = (mp_ptr) dp;
135
56.8k
    n2p = TMP_ALLOC_LIMBS (nn + 1);
136
56.8k
    MPN_COPY (n2p, np, nn);
137
56.8k
    n2p[nn] = 0;
138
56.8k
    nn += adjust;
139
56.8k
        }
140
141
61.9k
      invert_pi1 (dinv, d2p[dn - 1], d2p[dn - 2]);
142
61.9k
      if (BELOW_THRESHOLD (dn, DC_DIV_QR_THRESHOLD))
143
61.7k
        mpn_sbpi1_div_qr (qp, n2p, nn, d2p, dn, dinv.inv32);
144
257
      else if (BELOW_THRESHOLD (dn, MUPI_DIV_QR_THRESHOLD) ||   /* fast condition */
145
0
         BELOW_THRESHOLD (nn, 2 * MU_DIV_QR_THRESHOLD) || /* fast condition */
146
0
         (double) (2 * (MU_DIV_QR_THRESHOLD - MUPI_DIV_QR_THRESHOLD)) * dn /* slow... */
147
0
         + (double) MUPI_DIV_QR_THRESHOLD * nn > (double) dn * nn)    /* ...condition */
148
257
        mpn_dcpi1_div_qr (qp, n2p, nn, d2p, dn, &dinv);
149
0
      else
150
0
        {
151
0
    mp_size_t itch = mpn_mu_div_qr_itch (nn, dn, 0);
152
0
    mp_ptr scratch = TMP_ALLOC_LIMBS (itch);
153
0
    mpn_mu_div_qr (qp, rp, n2p, nn, d2p, dn, scratch);
154
0
    n2p = rp;
155
0
        }
156
157
61.9k
      if (cnt != 0)
158
5.15k
        mpn_rshift (rp, n2p, dn, cnt);
159
56.8k
      else
160
56.8k
        MPN_COPY (rp, n2p, dn);
161
61.9k
      TMP_FREE;
162
61.9k
      return;
163
61.9k
    }
164
165
  /* When we come here, the numerator/partial remainder is less
166
     than twice the size of the denominator.  */
167
168
36.4k
    {
169
      /* Problem:
170
171
         Divide a numerator N with nn limbs by a denominator D with dn
172
         limbs forming a quotient of qn=nn-dn+1 limbs.  When qn is small
173
         compared to dn, conventional division algorithms perform poorly.
174
         We want an algorithm that has an expected running time that is
175
         dependent only on qn.
176
177
         Algorithm (very informally stated):
178
179
         1) Divide the 2 x qn most significant limbs from the numerator
180
      by the qn most significant limbs from the denominator.  Call
181
      the result qest.  This is either the correct quotient, but
182
      might be 1 or 2 too large.  Compute the remainder from the
183
      division.  (This step is implemented by an mpn_divrem call.)
184
185
         2) Is the most significant limb from the remainder < p, where p
186
      is the product of the most significant limb from the quotient
187
      and the next(d)?  (Next(d) denotes the next ignored limb from
188
      the denominator.)  If it is, decrement qest, and adjust the
189
      remainder accordingly.
190
191
         3) Is the remainder >= qest?  If it is, qest is the desired
192
      quotient.  The algorithm terminates.
193
194
         4) Subtract qest x next(d) from the remainder.  If there is
195
      borrow out, decrement qest, and adjust the remainder
196
      accordingly.
197
198
         5) Skip one word from the denominator (i.e., let next(d) denote
199
      the next less significant limb.  */
200
201
36.4k
      mp_size_t qn;
202
36.4k
      mp_ptr n2p, d2p;
203
36.4k
      mp_ptr tp;
204
36.4k
      mp_limb_t cy;
205
36.4k
      mp_size_t in, rn;
206
36.4k
      mp_limb_t quotient_too_large;
207
36.4k
      unsigned int cnt;
208
209
36.4k
      qn = nn - dn;
210
36.4k
      qp[qn] = 0;       /* zero high quotient limb */
211
36.4k
      qn += adjust;     /* qn cannot become bigger */
212
213
36.4k
      if (qn == 0)
214
247
        {
215
247
    MPN_COPY (rp, np, dn);
216
247
    TMP_FREE;
217
247
    return;
218
247
        }
219
220
36.1k
      in = dn - qn;   /* (at least partially) ignored # of limbs in ops */
221
      /* Normalize denominator by shifting it to the left such that its
222
         most significant bit is set.  Then shift the numerator the same
223
         amount, to mathematically preserve quotient.  */
224
36.1k
      if ((dp[dn - 1] & GMP_NUMB_HIGHBIT) == 0)
225
24.1k
        {
226
24.1k
    count_leading_zeros (cnt, dp[dn - 1]);
227
24.1k
    cnt -= GMP_NAIL_BITS;
228
229
24.1k
    d2p = TMP_ALLOC_LIMBS (qn);
230
24.1k
    mpn_lshift (d2p, dp + in, qn, cnt);
231
24.1k
    d2p[0] |= dp[in - 1] >> (GMP_NUMB_BITS - cnt);
232
233
24.1k
    n2p = TMP_ALLOC_LIMBS (2 * qn + 1);
234
24.1k
    cy = mpn_lshift (n2p, np + nn - 2 * qn, 2 * qn, cnt);
235
24.1k
    if (adjust)
236
13.5k
      {
237
13.5k
        n2p[2 * qn] = cy;
238
13.5k
        n2p++;
239
13.5k
      }
240
10.6k
    else
241
10.6k
      {
242
10.6k
        n2p[0] |= np[nn - 2 * qn - 1] >> (GMP_NUMB_BITS - cnt);
243
10.6k
      }
244
24.1k
        }
245
12.0k
      else
246
12.0k
        {
247
12.0k
    cnt = 0;
248
12.0k
    d2p = (mp_ptr) dp + in;
249
250
12.0k
    n2p = TMP_ALLOC_LIMBS (2 * qn + 1);
251
12.0k
    MPN_COPY (n2p, np + nn - 2 * qn, 2 * qn);
252
12.0k
    if (adjust)
253
1.08k
      {
254
1.08k
        n2p[2 * qn] = 0;
255
1.08k
        n2p++;
256
1.08k
      }
257
12.0k
        }
258
259
      /* Get an approximate quotient using the extracted operands.  */
260
36.1k
      if (qn == 1)
261
8.76k
        {
262
8.76k
    mp_limb_t q0, r0;
263
8.76k
    udiv_qrnnd (q0, r0, n2p[1], n2p[0] << GMP_NAIL_BITS, d2p[0] << GMP_NAIL_BITS);
264
8.76k
    n2p[0] = r0 >> GMP_NAIL_BITS;
265
8.76k
    qp[0] = q0;
266
8.76k
        }
267
27.4k
      else if (qn == 2)
268
8.78k
        mpn_divrem_2 (qp, 0L, n2p, 4L, d2p); /* FIXME: obsolete function */
269
18.6k
      else
270
18.6k
        {
271
18.6k
    invert_pi1 (dinv, d2p[qn - 1], d2p[qn - 2]);
272
18.6k
    if (BELOW_THRESHOLD (qn, DC_DIV_QR_THRESHOLD))
273
17.5k
      mpn_sbpi1_div_qr (qp, n2p, 2 * qn, d2p, qn, dinv.inv32);
274
1.05k
    else if (BELOW_THRESHOLD (qn, MU_DIV_QR_THRESHOLD))
275
1.05k
      mpn_dcpi1_div_qr (qp, n2p, 2 * qn, d2p, qn, &dinv);
276
0
    else
277
0
      {
278
0
        mp_size_t itch = mpn_mu_div_qr_itch (2 * qn, qn, 0);
279
0
        mp_ptr scratch = TMP_ALLOC_LIMBS (itch);
280
0
        mp_ptr r2p = rp;
281
0
        if (np == r2p) /* If N and R share space, put ... */
282
0
          r2p += nn - qn; /* intermediate remainder at N's upper end. */
283
0
        mpn_mu_div_qr (qp, r2p, n2p, 2 * qn, d2p, qn, scratch);
284
0
        MPN_COPY (n2p, r2p, qn);
285
0
      }
286
18.6k
        }
287
288
36.1k
      rn = qn;
289
      /* Multiply the first ignored divisor limb by the most significant
290
         quotient limb.  If that product is > the partial remainder's
291
         most significant limb, we know the quotient is too large.  This
292
         test quickly catches most cases where the quotient is too large;
293
         it catches all cases where the quotient is 2 too large.  */
294
36.1k
      {
295
36.1k
        mp_limb_t dl, x;
296
36.1k
        mp_limb_t h, dummy;
297
298
36.1k
        if (in - 2 < 0)
299
1.96k
    dl = 0;
300
34.2k
        else
301
34.2k
    dl = dp[in - 2];
302
303
36.1k
#if GMP_NAIL_BITS == 0
304
36.1k
        x = (dp[in - 1] << cnt) | ((dl >> 1) >> ((~cnt) % GMP_LIMB_BITS));
305
#else
306
        x = (dp[in - 1] << cnt) & GMP_NUMB_MASK;
307
        if (cnt != 0)
308
    x |= dl >> (GMP_NUMB_BITS - cnt);
309
#endif
310
36.1k
        umul_ppmm (h, dummy, x, qp[qn - 1] << GMP_NAIL_BITS);
311
312
36.1k
        if (n2p[qn - 1] < h)
313
2.51k
    {
314
2.51k
      mp_limb_t cy;
315
316
2.51k
      mpn_decr_u (qp, (mp_limb_t) 1);
317
2.51k
      cy = mpn_add_n (n2p, n2p, d2p, qn);
318
2.51k
      if (cy)
319
981
        {
320
          /* The partial remainder is safely large.  */
321
981
          n2p[qn] = cy;
322
981
          ++rn;
323
981
        }
324
2.51k
    }
325
36.1k
      }
326
327
36.1k
      quotient_too_large = 0;
328
36.1k
      if (cnt != 0)
329
24.1k
        {
330
24.1k
    mp_limb_t cy1, cy2;
331
332
    /* Append partially used numerator limb to partial remainder.  */
333
24.1k
    cy1 = mpn_lshift (n2p, n2p, rn, GMP_NUMB_BITS - cnt);
334
24.1k
    n2p[0] |= np[in - 1] & (GMP_NUMB_MASK >> cnt);
335
336
    /* Update partial remainder with partially used divisor limb.  */
337
24.1k
    cy2 = mpn_submul_1 (n2p, qp, qn, dp[in - 1] & (GMP_NUMB_MASK >> cnt));
338
24.1k
    if (qn != rn)
339
551
      {
340
551
        ASSERT_ALWAYS (n2p[qn] >= cy2);
341
551
        n2p[qn] -= cy2;
342
551
      }
343
23.6k
    else
344
23.6k
      {
345
23.6k
        n2p[qn] = cy1 - cy2; /* & GMP_NUMB_MASK; */
346
347
23.6k
        quotient_too_large = (cy1 < cy2);
348
23.6k
        ++rn;
349
23.6k
      }
350
24.1k
    --in;
351
24.1k
        }
352
      /* True: partial remainder now is neutral, i.e., it is not shifted up.  */
353
354
36.1k
      tp = TMP_ALLOC_LIMBS (dn);
355
356
36.1k
      if (in < qn)
357
11.3k
        {
358
11.3k
    if (in == 0)
359
1.62k
      {
360
1.62k
        MPN_COPY (rp, n2p, rn);
361
1.62k
        ASSERT_ALWAYS (rn == dn);
362
1.62k
        goto foo;
363
1.62k
      }
364
9.73k
    mpn_mul (tp, qp, qn, dp, in);
365
9.73k
        }
366
24.8k
      else
367
24.8k
        mpn_mul (tp, dp, in, qp, qn);
368
369
34.5k
      cy = mpn_sub (n2p, n2p, rn, tp + in, qn);
370
34.5k
      MPN_COPY (rp + in, n2p, dn - in);
371
34.5k
      quotient_too_large |= cy;
372
34.5k
      cy = mpn_sub_n (rp, np, tp, in);
373
34.5k
      cy = mpn_sub_1 (rp + in, rp + in, rn, cy);
374
34.5k
      quotient_too_large |= cy;
375
36.1k
    foo:
376
36.1k
      if (quotient_too_large)
377
1.64k
        {
378
1.64k
    mpn_decr_u (qp, (mp_limb_t) 1);
379
1.64k
    mpn_add_n (rp, rp, dp, dn);
380
1.64k
        }
381
36.1k
    }
382
36.1k
  TMP_FREE;
383
36.1k
  return;
384
34.5k
      }
385
110k
    }
386
110k
}